21 replies to this topic

[bragadocchio]

Bookmark Managers the Key to Personalized Search

A link popularity method of indexing web pages is supposed to be about votes for "important" pages on the web. But there's one set of votes that this type of measurement misses, which may be an even greater measure of the importance of sites - bookmarks. Important, at least, to the people visiting those pages.

A new patent application from Googlers Gregory Joseph Badros and Stephen R. Lawrence describes Methods and systems for personalized network searching. The application provides some background on why bookmarks might be helpful to searchers who might want personalized searches.

They begin by telling us that most web page visits are actually revisits, where a user is returning to a previously visited page. When people go to those pages, they often use conventional search engines, instead of typing in URLs in their browser's address bar or using browser bookmarks. They note that these searches are navigational rather than ones using keywords to find information, and searchers are using navigational searches more frequently (See Understanding User Goals in Web Search for more on navigational searches.)

But, bookmarks are still being used by many, especially when a search engines doesn't rank the site highly or it is hard to find with a search query. The inventors tell us that bookmarks that are used frequently can be a valuable resource for the user. In the context of this patent application, they mean more with this statement than just as a convenient way to return to favorite pages.

An Improved Bookmark Manager is Your Friend

There are some problems with the bookmark methods that most browsers use. For instance, people often use more than one computer, at home and work, and it can be difficult to share those bookmarks between those places, even though the bookmarks might be useful at both. It's also not always easy to organize bookmarks, which can make it difficult for the user to find a favorite site.

There are browser-based and commercial bookmark management tools some people use, and some of them can make synchronization between computers easier, but often fail to address the organizational problems of bookmarks. Other tools make it easier to use bookmarks, such as automatically organizing bookmarks on a computer, searching older bookmarks by keyword, and integrating the back, history, and bookmark functions to improve the user's ability to visit previously visited sites (see, e.g., Integrating Back, History and Bookmarks in Web Browsers).

One place where these tools fail users is that they don't have the ability to provide personalized search results based upon bookmarks saved by the user.

That's the need that the technology described in this patent application aims to fill.

Bookmarks as Part of "Personalized Search Objects"

Personalized search may begin with the use of a bookmark manager, which makes it easy to create, modify, delete, and save bookmarks on the network. It would also provide information about those bookmarks to the search engine. The search engine can return searches based upon documents linked to in the bookmark manager alone. Those bookmarks can be shared over different computers. The manager may also include user ratings for different documents bookmarked. Multiple user personalities can be defined in the bookmark manager, with "recommendations" based upon those different personalities.

The bookmarks in the bookmark manager may be part of a "personalized search object" which could also include the history list of the browser. That personalized search could be combined with a general search on the web, or may display both types of searches side-by-side. A number of different ways to sort results are noted in the patent application, including sorting of the bookmarks by user ratings, and possibly even sorting of the general ratings based upon topics defined by the users bookmarks.


Bookmarks Build Communities

It may also be possible to develop a community set of bookmarks, that can be shared, and can help define search results.

The selection of bookmarks may also enable the search engine to define communities on the web who share similar bookmarks.


More than URLs Measured

Repeat visits to a site, amount of time lingering at a bookmarked page, and infrequent use of some bookmarks may also be measured, and inform the results of searches. Other measures, such as whether a page is printed, saved, or scrolled upon could also be noted. A user-defined bookmark title may also influence the search, as may user ratings and annotations of bookmarked sites.

Existing favorites from a browser can be migrated to the bookmark manager program. The hierarchy of those migrated bookmarks (the way they are organized) can be used as annotations for the URLs, as can the full text of an article linked to, when the page was last visited. As browser based bookmarks are edited, the ones in the bookmark manager can be, too.

An html interface might be used to allow a person to change ratings for bookmarked sites, and to add or change annotations. It may also show recently visited, and recently rated sites - enabling a "work-list like review of a surfing session."

The patent application describes the possibility of a standardized format for annotations, as described in Annotea: An Open RDF Infrastructure for Shared Web Annotations


How it Works

A search can combine three processes: searching global indices on the web, searching the bookmarks, and searching annotations of the bookmarks (and possibly also searching the "navigation history" of the browser).

Searching annotations comprises searching the user-entered annotations using the search query 114 submitted by the user 112a. For example, a user 112a may enter the term "boat" as an annotation for a page comprising marine supplies. If the users 112a enters "boat" as part of the search query 114 utilized by the search engine 120, the page with the "boat" annotation will be returned by the search annotations component. Another embodiment of the present invention searches not only the pages that the user has bookmarked or annotated, but also pages similar to the pages that the user has bookmarked or pages with similar annotations.

Annotations that are created by one user may be used by "other users who share similar interests with the user who provided the annotation."

The text of annotations may be used as "user-specific anchor text" referring to the annotated URL.

Annotation may also provide shortcuts to pages in navigational type searches, by allowing a searcher to associate something like the word "home" to the corporate home page. A search for "home" would then bring them to that page.

The application describes a number of different types of interfaces, and their features, that can be used, including a simple one, and intermediate level, and an advanced interface.


Convincing the Unwashed Masses

Liked the honesty of these statements in the patent application, which map out potential marketing methods for the adoption of this type of bookmark manager:

Embodiments of the present invention implement various measures to help encourage user adoption. For example, although not all users may be willing to expend the effort to provide ratings, an embodiment of the present invention provides noticeable benefits for relatively low effort on the part of the user. In addition, by incorporating bookmark synchronization, an embodiment of the present invention helps drive adoption.

Embodiments of the present invention may also implement network and community features to foster adoption of the service. For example, as described above, an embodiment of the present invention may utilize like-user recommendations to locate and rank results. One embodiment of the present invention implements user groups and friend-lists whereby a user can choose to expose a bookmark list to friends or the public at large. In another embodiment, a user has the ability to transparently overlay a weighted set of bookmarks onto their own set of bookmarks.

An organization implementing an embodiment of the present invention may utilize partnerships to encourage adoption of the service. For example, a service provider may encourage partner sites to display a "bookmark this page!" snippet on their homepages and other content pages. For the partner, an embodiment of the present invention provides a means to ask users to opt-in to making it especially easy to get at their site via a search. And for users it's a nice reminder to mark the page or add an annotation. For the provider of the bookmark and search service, such an arrangement helps introduce users to the idea of bookmarks at the moment it matters most: when they are visiting a page they are interested in. It may be advantageous to (e.g., for security reasons) to have partners wishing to display a "bookmark this page!" link to register with the service provider first. Registration with the service provider also helps the service provider to develop relationships with additional content providers.

A provider of a bookmark service may receive various benefits from implementing the service. For example, the provider is able to collect data concerning users' attribution of value on pages.

Beating Spam

Antispam methods of the system may include only local implementation (as opposed to global), credit card validation, or CAPTCHAs (Completely Automated Public Turing Test to Tell Computers and Humans Apart).

It may also include "collaborative link recommendations" (Searchers who visited this page, and bookmarked some of the same sites as you, also visited this page...)


Pattern recognition, Better Ads, Rating People's Bookmarks, and Social Networking

By tracking the visits of users of the bookmark manager to different sites, and at different times of the day, the:

service provider has access to the previously unavailable data and may be better equipped to provide user-personalized portals. For example, pattern recognition might let the service provider realize that a user visit various stock quotes every Monday morning, checks CNN.com in the afternoons, etc. In such an embodiment, the search engine 120 may anticipate the pages that users will likely require.

Allowing the search engine to manage your bookmarks may also increase the relevancy of the advertisements presented to you when performing searches. It can tie together your interests as defined by your bookmarks with click-throughs of ads by searchers with similar interests:

This feature provides numerous benefits. Not only are users more likely to be satisfied because the advertising is more targeted, but the click-through rate for the service provider may increase, resulting in increased revenue.

A user may share bookmarks with others, or incorporate bookmarks from others into their own. An example given:

"Add this group's bookmarks to your favorites" when joining a new mailing list.

This was an interesting set of statements regarding ranking pages:

In one such embodiment, the bookmark indicators in results pages distinguish between those pages explicitly bookmarked by the user from those gathered by others. Given a canonical URL through which to reference another individual/organization's bookmarks, the service provider can derive a sense of the popularity of a person's links and weight those bookmarks correspondingly (a la PageRank applied to the subgraph of bookmark interlinks).

The potential applications for this type of bookmark manager seem unlimited, with even a potential social networking feature addressed:

One embodiment of the present invention fosters community and relationship building. In one embodiment, the search engine is able to recognize clusters or pairs of users having similar interests. Such an embodiment is able to suggest other users with which to network.

Adding Bookmarks, and Filtering Sites, without User Intervention

The bookmark manager may also bookmark pages that you don't add to your bookmarks, by looking at such things as "linger time and/or repeat visits." A positive rating for a page may happen if you spend a lot of time at a page (with a potential user override ability for that rating.) A rating score system is described in the patent application (guess that they couldn't write a patent application without throwing some math into it.)


Negative ratings could be used to exclude search results shown to users.


Mitigating Privacy Concerns


Of course, any privacy concerns that we might have are easily addressed (sarcasm intended here on my part):

To mitigate privacy concerns, embodiments of the present invention may require users to opt-in to the tracking. In such an embodiment, the system alerts the user when personalized search is in effect and provides a simple mechanism for reverting to generic search. In such an embodiment, bookmark data may be stored in a secure data center separate from a user's other personal data.

Costs and Benefits


I'd wrap this up with a conclusion, but I don't know if I could match the cold calculus of the patent application in its cost benefit analysis of why people might adopt a bookmark manager issued by the search engine:

Embodiments of the present invention provide numerous advantages to the user and to the provider of the search service. An embodiment of the present invention improves the user experience by providing personalized search results and rankings. An embodiment of the present invention provides advantages to the provider of a search service by (1) increasing the stickiness of the search experience by giving users a compelling reason to identify themselves and share their interest in topics with the provider, and (2) gathering better data regarding the relevancy of pages to different users and different classes of users.

In an embodiment of the present invention, the user providing bookmarks to the service provider enables the search provider to personalize the search for them. The feature can be viewed as a server-side generalization of bookmarks integrated with annotations. Users are able to share that personalization data across different browsers (e.g., work and home) if desired and hence eliminate the drudgery associated with managing bookmarks. An embodiment of the present invention also unifies all navigational queries under a single experience.

Thank God. My drudgery can now come to an end. (Do I need to note my sarcasm again?)

[whitemark]

When it comes to using Google, I have various emotional conflicts. On one hand is the temptation to use that very useful free tool and on the other is the concern over the invasion of my privacy. Then there's the 'Do No Evil' admiration of the brand, and yet, I have no idea what information is actually deleted (if any) when I click "Delete all account info" in my Google Account. (The wordings are of course reassuring and gives you an excuse to continue using Google :).

[bragadocchio]

I have the same conflicting emotions.

There are a lot of potentially useful applications of a tool like this. I could see building a collection of bookmarks for the forum, for instance. But the idea of Google tracking and collecting information on everytime people visit one of the pages in it, clocking how long they stay there, and even whether they scrolled on the page would make me hesitate.

I could see it helping me find information that is similar to stuff I have in my bookmarks, but I'm not sure any of that information is any of their business. It wasn't until I read this patent application that it really struck me why I wouldn't want Google to have its own browser.

I can even envision Google returning much more relevant results if everyone used a bookmark manager like this. But I don't know if I want to pay that cost.

[bwelford]

... and even whether they scrolled on the page

How would they know that, Bill, if you don't click on anything 'below the fold'?

[bragadocchio]

That's a good question, Barry. Here's the paragraph from the Patent Application where they mention that metric:

A user 112a may specify bookmarks explicitly. In one embodiment, the bookmarks are implicitly identified based on a measure of the behavior of the user. For example, in one embodiment, the implicit measure comprises the linger time. In other words, if a user spends a great deal of time on a site, it is identified as a bookmark for later personalized searches. In other embodiments, the implicit measure may comprise at least one of the quantity of repeat visits to the site or the quantity of click-throughs on the site. In one embodiment, temporal decay of ratings may be utilized so that unused or rarely used bookmarks, whether explicitly marked or implicitly marked, become unmarked over time. Other implicit measures include printing the page, saving the page, and the amount of scrolling performed on the page.*

*My emphasis.

They mention earlier in the patent application the ability to update the Bookmark Manager if you add bookmarks to your browsers favorites. Their integration with a browser might be deeper than that if they are tracking whether or not you scroll on a page, or using a print button, or spend any amount of time on a page.

The folks at Google, back when they were denying the creation of a Google browser, did mention that they were looking seriously at "browser enhancements." This type of user tracking sounds like an enhancement to me. But one that I'm not sure people might like.

[bragadocchio]

A better answer.

All it may take, Barry, is a simple java script.

Here's a description of how to measure how much someone has scrolled on a page with a working example:

http://www.howtocrea...p?tut=0&part=16

Scroll down to: "Finding how far the window has been scrolled"

This page describes a similar approach:

Browser Detecting (and what to do Instead)

Also see this paper on a Curious browser (pdf), which collects information on mouse clicks and movements, scrolling time and clicks on the scroll bar, use of keyboard pageup and pagedown, and collects ratings.

[gravelsack]

hmmmm... I have a feeling there are going to be a bunch of spammers signing up for this and bookmarking a handful of authority sites, and a bunch of their spam stuff. Probably many times over using different names - just in case it becomes a major factor in the algo.

.

[Eddie]

I have no idea what information is actually deleted (if any) when I click "Delete all account info" in my Google Account.

I beleive that nothing is deleted at all. It all stays on googles servers.

Does anyone really still swallow the do no evil motto ?

Google is a very large business, not an organisation for doing good, just to make money.

[whitemark]

Does anyone really still swallow the do no evil motto ?

Actually, yes, I do.
It might sound like an idealist's vision, but the people behind Google are not the kind of idealist who just fantasize about their idealism, and yet not do anything about it in reality. They are the kind of idealist, like Martin Luther King or Gandhi, who actually act on their idealism because they truly believe in it.

Their history is testimony to this - These people started out with grit, determination and an idea nobody but them believed in and despite great odds, established a $80 billion company. And despite going public, they still have managed to retain great control and clout over their board. (Its when the founders are chucked out, that I'd really start worrying.)

Unfortunately there is so much cynicism in todays world, that even 'established' principles - like Gandhi's idea of non-violence - are considered an idealistic utopian fantasy. When Gandhi himself is doubted in today's world, is it any wonder that people doubt the 'do good' motives of others!

----------------------------------------
My conflicts of emotions are not over how they intend to use my data, but the level of privacy I am comfortable sharing with them. When I find myself saying that I am not comfortable with Google knowing this about me, it really makes me pause and think about my actions or thoughts.

[bwelford]

Those are excellent sentiments, whitemark. I tend to be at your end of the scale although not quite as far along.

For me, the question relates to who is in control of the relationship. If Google is controlling the relationship, then I'm not happy. If I control the relationship, then I am.

[Eddie]

They are the kind of idealist, like Martin Luther King or Gandhi, who actually act on their idealism because they truly believe in it.  

I just can't swallow that myself. They have done nothing like any of these people. All they have done is start a technology company, and as soon as they could they got every penny they could by monetising every part of the business.

Look at all the way they want to place links on your site. When Microsoft did this they were the evil empire.

You are of course entitled to your opinions, but to my mind comparing google to Ghandi is an insult of the highest order to Ghandi.

[bwelford]

You're right, Eddie, of course. They were never in the same ballpark as folk such as Ghandi. However in the early days I think there may have been a certain idealism on the part of Page and Brin to deliver the best benefits to searchers. I don't think that slogan "Do no evil." was just empty cynicism. However I think there is a tendency for very bright people to think they can do good things for you, beyond what you think you need. That may explain the early tendency to go beyond what many thought reasonable.

Now of course, as a public company with shareholders, they must deliver financial results and that produces a different balance between user benefits and cash generation.

[whitemark]

I tend to be at your end of the scale although not quite as far along ...

It's one of the things I admire about Bill and you - the balanced views. I sometime see things only in black or white, to my detriment.

They have done nothing like any of these people. All they have done is start a technology company ...

The comparison does sound a bit off because of Gandhi's greatness. But what I am comparing is the idealism of the Mahatma and the Google founders (as Barry rightly pointed out).

Doesn't 'Do no evil' sound idealistic; as the fantasy of a young man who has no idea about the real world? Well, so do Gandhi's principles of truth and non-violence. Even during his time some considered him a fraud -

However I think there is a tendency for very bright people to think they can do good things for you, beyond what you think you need.

So very true. One wonders whether Google's idealism will eclipse its reasoning ...

[Eddie]

You may be right that they started out with the best of intentions - who knows, but in the end you are judged by your actions, and imo google don't come out that well.

[bragadocchio]

hmmmm... I have a feeling there are going to be a bunch of spammers signing up for this and bookmarking a handful of authority sites, and a bunch of their spam stuff. Probably many times over using different names - just in case it becomes a major factor in the algo.

That was one of the thoughts that entered my mind, too.

There was a section of Google's Historic Data patent application that didn't seem to make much sense until this newer patent application came out:

[0114] According to an implementation consistent with the principles of the invention, user maintained or generated data may be used to generate (or alter) a score associated with a document. For example, search engine 125 may monitor data maintained or generated by a user, such as "bookmarks," "favorites," or other types of data that may provide some indication of documents favored by, or of interest to, the user. Search engine 125 may obtain this data either directly (e.g., via a browser assistant) or indirectly (e.g., via a browser). Search engine 125 may then analyze over time a number of bookmarks/favorites to which a document is associated to determine the importance of the document.  

[0115] Search engine 125 may also analyze upward and downward trends to add or remove the document (or more specifically, a path to the document) from the bookmarks/favorites lists, the rate at which the document is added to or removed from the bookmarks/favorites lists, and/or whether the document is added to, deleted from, or accessed through the bookmarks/favorites lists. If a number of users are adding a particular document to their bookmarks/favorites lists or often accessing the document through such lists over time, this may be considered an indication that the document is relatively important. On the other hand, if a number of users are decreasingly accessing a document indicated in their bookmarks/favorites list or are increasingly deleting/replacing the path to such document from their lists, this may be taken as an indication that the document is outdated, unpopular, etc. Search engine 125 may then score the documents accordingly.  

[0116] In an alternative implementation, other types of user data that may indicate an increase or decrease in user interest in a particular document over time may be used by search engine 125 to score the document. For example, the "temp" or cache files associated with users could be monitored by search engine 125 to identify whether there is an increase or decrease in a document being added over time. Similarly, cookies associated with a particular document might be monitored by search engine 125 to determine whether there is an upward or downward trend in interest in the document.

I've seen a few articles recently about the historical data patent application, pointing to this section as "proof" that Google is developing a browser. But, this bookmark manager program sounds more likely.

[rcjordan]

Bill, you've seem uncharacteristically cynical about G's motives ever since msgraph outed the patent and you did the detailed review --but perhaps it was even before that. Is the patent causing you concern? The tranformation of G to a "media company?" Other (mis)deeds? All of the above?

[bragadocchio]

I'm not sure that I meant to come across quite as cynical as I did there, RC. It just came out that way.

Funny, I didn't see msgraph's post about this patent at SEW until Monday. There are often between 5,000 and 7,000 patent applications that get published every week, and I've been making a habit of searching through them over the past few months, and keeping some notes.

Come across some interesting stuff that way, like this Dynamic content clustering patent application from Microsoft last week for news type information, Yahoo!'s application for Systems and methods for implementing person-to-person money exchange, Google's application for Generating user information for use in targeted advertising, and the personalized book mark manager application that I decided to spend a few hours digging through and writing about on Saturday morning.

There were also some great applications from Microsoft's UK research team on browser navigation enhancements (Intelligent backward resource navigation, Intelligent forward resource navigation , Browser session overview ), which were worked on, in part, by a team member who has since left for Google, and may have played a role in developing some of the ideas in the application we are discussing (some nice papers linked to on that site that describe some of the technology behind those patents, as does the Microsoft Integrated Systems Group page).

There is a lot of potential in Google's search information development to do good, and there's a lot of potential for it to cause harm. A corporation can be a responsible social entity, and by all respects, Google has so far shown themselves to be.

But the risk is there, as in most things, for that to change. Brin and Page do appear to be filled with idealism, and intent to make things better for people. But there's always the risk that the people whom things are being made better for can narrow, so that they focus more on the shareholders of a company than those whom a service is being provided for.

I guess when I wrote this, I was envisioning a Google where the type of information being shared with them was being put to uses that may not be as shiny and bright as we would like to believe would come from the company.

In the Yahoo! patent application above, one of the points behind the invention was to help people from having to send confidential information across the internet. In the Microsoft applications, the management of book marks and navigation they describe seem like individual tools that help the user, without some central repository of data that can be mined for any reason whatsoever.

Google is a private company, but with patent applications like this one, and the historical data patent, there is the potential to share a lot of information that we really may not want to share, and have them create rankings of sites, and decide what we see when we search, based upon their own interpretations of our "user defined data." Cynicism, no, but concern, yes. Part of this patent application comes off to me as if the author was saying; "we have this great tool for you all, and all it will cost you is sharing everything you do on the web with us, so that we can use it however we want, including feeding you the ads that we know you want to see."

A social bookmarks manager like del.icio.us doesn't seem to hold the same potential risk. It's tools like that, that don't try to "organize the world's information" that keep me from falling into cynicism.

[rcjordan]

>when I wrote this, I was envisioning a Google where the type of information being shared with them was being put to uses that may not be as shiny and bright as we would like to believe

It's been my experience that cynicism is a far healthier attitude than optimism and/or blind faith, Bill. Contingency plans (and bunkers) aren't built by those that expect the best motives to prevail.

As for the information being collected, "personal information is the currency of the internet" and I think it's the sheer raw value of this commodity that gives you & I ample reason for concern regarding who's brokering the exchange.

>organize the world's information

HHH! Google's new mission: organize the world's money

[MrJones]

A lot of this talk about doing some interesting things with bookmarks reminded me of a past coworker who now works for SearchFox.
At first it was just a web based tool for sharing bookmarks, but now they added many of the features mentioned above including real-time searching of bookmarks and a browser toolbar.

I think the company is more of an acquisition play since I don't see how a user would pay for the service.

[swainzy]

It looks like Gmail has already had a security breach in Oct. of 2004.
Here is an article from
googlewebloginc.com

This article refers to an interview in the Israeli publication NetLife.

Hopefully this isn't old news to most or off subject.
Swainzy